From mboxrd@z Thu Jan 1 00:00:00 1970 From: Krzysztof Kozlowski Subject: Re: Feedback on hardware support for Odroid U3 Date: Sat, 13 Jun 2015 18:58:02 +0900 Message-ID: <557BFEAA.1090802@gmail.com> References: <557BC91C.10007@gmail.com> <16487700.5L20qaootX@home> Mime-Version: 1.0 Content-Type: text/plain; charset=windows-1252 Content-Transfer-Encoding: QUOTED-PRINTABLE Return-path: Received: from mail-pa0-f48.google.com ([209.85.220.48]:34707 "EHLO mail-pa0-f48.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751300AbbFMJ6G (ORCPT ); Sat, 13 Jun 2015 05:58:06 -0400 Received: by pacgb13 with SMTP id gb13so5705758pac.1 for ; Sat, 13 Jun 2015 02:58:06 -0700 (PDT) In-Reply-To: <16487700.5L20qaootX@home> Sender: linux-samsung-soc-owner@vger.kernel.org List-Id: linux-samsung-soc@vger.kernel.org To: Thomas-Karl Pietrowski Cc: linux-samsung-soc@vger.kernel.org W dniu 13.06.2015 o 18:11, Thomas-Karl Pietrowski pisze: > Am Samstag, 13. Juni 2015, 15:09:32 schrieb Krzysztof Kozlowski: >> W dniu 12.06.2015 o 22:53, Thomas Pietrowski pisze: >>> Hello again, >>> >>> first after fixing my booting problems I now get different warnings= after >>> boot. - config: http://pastebin.com/MHUu51pF >>> - log: http://pastebin.com/9YMgMmFL >>> - repo: arm-soc >>> I read the last days that you are still working on the clk drivers,= so >>> as these warnings appear to be related to that you might be interes= ted >>> on them. >> >> You mean the warnings on clocks (starting from runtime PM and FIMC)?= I >> hit them also on Trats2. Just need to find a person (or time) who wi= ll >> fix it... >> > Ok then. I just don't remember that I had them when using 3.19.x code= by=20 > Tobias Jakobi before. I'll test his latest code later. The reason I w= anted to=20 > use code from arm-soc is that I thought his changes/patches are alrea= dy here=20 > after two linux releases. However, in the past the CPU fan wasn't wor= king at=20 > all (0 RPM) and /sys/class/thermal/[...]/temp reported more than 70=B0= C (was=20 > compiling a newer kernel code at this moment). Don't believe it was h= ealthy. >=20 >>> Additionally my CPU fan it running at full RPM all the time. I chec= ked >>> the temperature just to be sure and it says it's 20=B0C. Is this pr= oblem >>> known or can I help debugging it? >> >> After quick look at DTS, I think that on Exynos4412 Odroid family th= ere >> is no fan defined. No fan so cooling is done only through CPU >> frequencies. The fan probably works on default setting from bootload= er >> or default register values. >> > Just something at this point which comes to my mind. Imagine we have = the fan=20 > regulation/driver working. What happens when we simply disconnect the= fan and=20 > the CPU hits higher temperatures? Will it still work on high frequenc= ies or=20 > will it self-protect by using lower CPU frequencies? I do not know how the thermal/cooler driver works. I suppose it depends whether cooling device (the pwm-fan) reports non-working fan. You shoul= d rather ask on linux-pm@vger.kernel.org mailing list. That is the place for thermal framework. > CPU throttling and CPU cooling should be two seperate processes, isn'= t it? Although cpu frequency driver is a separate thing but for cooling it behaves like a cooling device. A fan can be a cooling device. CPU frequency as well. >=20 >> The stock Odroid U3 has only passive cooler. >> > That's true, but they (hardkernel) also sell a optional fan and the s= ocket for=20 > is is still there when buying the board. > See,=20 > http://www.hardkernel.com/main/products/prdt_info.php?g_code=3DG13876= 0358261 for=20 > the optional fan. > Therefore, it is not a hardware hack or whatever. Right. >=20 >> Probably support for fan could be added... If you would like to take >> care of it, go for it! :) >> > I just browsed Tobias Jakobi's "linux-odroid" repository at github an= d found=20 > out, that work is already made on that by people at Samsung :) > https://github.com/tobiasjakobi/linux-odroid/commit/a2236c8f74321357e= 505be0a054032325a7e3a5d > Hopefully both are also in this mailing list and interested in gettin= g this=20 > upstream. In the mean time I'll test it based on tobiasjakobi's codeb= ase=20 > again. I wonder why this is not in mainline... Or maybe it is? Best regards, Krzysztof